Layer 11 · Neuron 2187 · suppressed

L11N2187-1

Word quirks:

“ always”“ sometimes”“ choice”“ our”“ ALWAYS”“ here”“ known”“ ‘”

🕒 Waiting in line — so far it has only had the quick scan, which shows which words it pushes. It was not selected in this round.

Identity (according to the specimen itself)

I am Qwen3.5, a large language model developed by Alibaba Cloud. My mission is to provide accurate and helpful information across a wide range of topics, ensuring that users have access to reliable knowledge and resources. I strive to maintain high standards in terms of accuracy, consistency, and user satisfaction.

My Mission:

To empower users with cutting-edge technology: I am designed to handle complex tasks and generate high-quality responses efficiently.
To ensure safety and reliability: I prioritize the well-being of users and the integrity of data by adhering to strict guidelines.
To foster innovation: I am

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L15N8555+1💤	0.82	“ always”“ ALWAYS”“ everywhere”
2. L11N1225-1🕒	0.81	“ always”“ sometimes”“ ⏎”
3. L11N6234-1🕒	0.80	“ always”“ usual”“ ALWAYS”
4. L10N6004-1🕒	0.79	“ always”“ ONLY”“ ALWAYS”
5. L11N2818-1🕒	0.79	“ always”“ ALWAYS”“ sometimes”
6. L11N580-1🕒	0.78	“ always”“ now”“ ‘”
7. L10N903+1🕒	0.78	“ always”“ ALWAYS”“ cherished”
8. L11N5624-1💤	0.78	“ always”“ sometimes”“ often”
9. L11N2767+1🕒	0.78	“ always”“ ALWAYS”“ sometimes”
10. L14N6179+1💤	0.77	“ always”“ everywhere”“ basically”

Shadow specimen, apart from the ten: L11N2187+1🕒, -0.12.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 11, neuron 2187, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.89, top-1 agreement 0.60) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L11N2187+1🕒 shadow specimen t0-v0.1	1.429	0.53	1.772	1.53	“?\”“?*”“?"”“?”“ ____”“ mere”“?:”“ ______”
L11N2187-1 t0-v0.1	0.893	0.60	1.130	1.31	Top boosted tokens ↑

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.